Tuesday, September 19, 2006

The Global Warming Shell Game

Warmer: "Oh noes! We're all gonna DIE unless we change our ways RIGHT NOW! I've got a study here, Study A, that conclusively proves this beyond a shadow of a doubt!"
Skeptic: "Where's the data behind that conclusion?"
W: "You don't need to see my data. You'd misuse it."
S: (whines about data access to the journal, the media, the funding agency)
W: (grudgingly) "Oh, all right. here's the data."
S: "Hey, it looks like you still left out some data, and your method is broken."
W: "No, I didn't, you industry shill. And there's nothing wrong with
my methods."
S: "I'm pretty sure you did, and there is."
W: "There's nothing wrong."
S: "Yes there is."
W: "No there isn't."
S: "Yes."
W: No."
(Skeptic finally convinces an independent expert panel to review the situation.)
EXPERT PANEL: "S is right. W didn't make his data or methods fully available and as far as we could figure out from what he DID make available, his methods are suspect and conclusions totally unreliable."
Warmer: "Okay, sure, even if there /were/ issues with /that/ study, that's an eight-year-old study! Science doesn't stand still, you know!
Even if there were teensy problems with the methodology, the essential conclusions are still correct and this has been proven CONCLUSIVELY by Studies B and C, that will be published this summer. So oh noes! We're all gonna DIE unless we change our ways RIGHT NOW! I've got PROOF!"
Skeptic: "Studies B and C seem like they have a lot of the same problems as A did."
W: "No they don't, you industry shill."
S: Yes they do!"
....and so on.


At 10:37 PM, Blogger Brian said...

Nice fable, Glen.

You posted on my website that you'd be interested in betting me over global warming at the longbets site. I think we can make it happen. Please reply here or send me an email.

-Brian Schmidt, from Backseatdriving blog

At 1:01 AM, Blogger Glen said...

The problem with an even-odds bet is that in the short run the IPCC position has too wide a range. A pure linear projection says the planet warmed 1 degree in the last century so (assuming there's some momentum at work) it might be expected to warm another 1 degree in the next century. I'd be happy to bet even odds we're on the low side of that. Scale it down to a mere 20 years and we get .2 degrees expected warming.

The nice thing about Lindzen's bet offers is that it revealed how vague the warmers previously had been when it comes to confidence intervals and got them to retreat substantially. Before any talk of betting, there were claims of an extremely high confidence level, that the "consensus" was that temperatures would absolutely be in the range given. I perceive Annan's RC post as serious backpedaling in the direction of statistical sanity. What does it really mean to say you expect .3 degrees +/-0.1? If you're highly confident that range encompasses the likely values, we might call the range you give the 95% confidence interval. (2 standard deviations, not one). Which would mean there's only a 2.5% chance you're outside that interval on the LOW end. In which case it really would be appropriate for the warmer to offer the sort of extreme odds Lindzen originally proposed (in this case, 40:1). On the other hand, if we call 0.1 a single standard deviation then about 2/3rds of the time you're in the range and 1/6th of the time you're on the low end; the warmer should then offer 6:1 odds on a mere over/under.

From my perspective, the "IPCC doesn't know squat" perspective, .2 is a reasonable prediction; the "under" has something on the order of a 50% chance of being right. From your perspective, the "over" has a huge margin. If you really wanted to be fair and split the difference you'd have to go higher or do something like what Lindzen proposed for his second bet. But just for the entertainment value and essentially on a hunch, I'd be willing to bet on the "under" at even odds. For .2 degrees, $200.

However, I'm probably not the best person to take on the longbet. I have no reputational capital to maintain in this particular field. No degrees, very little expertise, no publications, no followers, no particular reason to care if anyone found out I was wrong. All I have is my skepticism and $200 to back it up.

Even stronger caveat, though: do we seriously believe longbet will be around in 20 years? What odds would you give me on that bet? :-)


Post a Comment

<< Home